THE PRAKRITS 31
closely akin to Sanskrit than Maharastrl ; its place of origin was within the sphere of the strongest influence of Sanskrit, and it remained in specially close relation with it both in morphology, syntax, and vocabulary. Hence it was appropriately used-for persons of good position in the drama. MagadhI, on the other hand, was reserved for those of low rank, and, though tales 1 were composed in it, it was of comparatively minor importance. The Ndtyagdstra, perhaps in the third century A. D., enumerates other dramatic dialects (vibhasds) which are clearly of no real popular origin ; such are Daksinatya, Pracya, AvantI, and Dhakkl or Takkl, which are mere varieties of (paurasenl, while Candali and (pakarl are species of MagadhI. 2 PaicacI, though practically unknown in the extant dramas, enjoyed, it appears, a consider- able vogue in the popular tale, as a result, doubtless, of the fame of the Brhatkathd.
The comparatively late date at which Maharastrl appears to have come into fame, as indicated by its exclusion until late from the drama, suggests that some other Prakrit was employed for poetry before its rise into repute. Jacobi has found traces of such a Prakrit in the verses cited in the Ndtyagdstra ; 3 it was marked by the facultative retention or change or loss of inter- vocalic consonants, and was akin on the one hand to (paurasenl, for example in such forms as sadisa for sadrga and the gerund in iya, while it shared with Maharastrl the locative in amnti and the gerund in una; from these local indications he suggests that it had its centre in Ujjayim. It was, he holds, from this dialect that the softening of t to d passed into (paurasenl, which in Afvaghosa hardly shows any trace of it, and also in the dialect, otherwise similar to Jain Maharastrl, which on this account Pischel 4 named Jain (paurasenl. This poetic Prakrit, like (paurasenl, is essentially closely akin to Sanskrit.
1 Probably in verse, like Maharasfii and Apabhranca tales ; Dandin, i. 38; Rudrata, xvi. 36. Dandin's Gaud! Prakrit may be MagadhI ; he mentions also Latl.
a Cf. Keith, Sanskrit Drama, pp. 140 ff., 337 ; Gawronski, KZ. xliv. 247 ff. Iranian traits in Cakarl aie not proved (JRAS. 1925, pp. 237 ff.) ; the points adduced all are essentially MagadhI (cf. ibid., pp. 218 ff.).
- Bhavisatta K'aha, pp. 84 ff. He does not touch on its relation to Pali
' Op. cit., § 21.
